Live-Demo Link: PRISMATIC
Web-deployed ML platform for computational genetic variant pathogenicity screening and repair
PRISMATIC is a full-stack ML application allowing users to input genetic mutation data in the following form:
Chromosome Number: integers 1-24 where X = 23 and Y = 24
Reference Sequence (Original DNA Allele) e.g. "ATCG"
Alternate Sequence (Variant DNA Allele) e.g. "ACAT"
Flank 1 ("Left" side of the alleles)
Flank 2 ("Right" side of the alleles)
PRISMATIC will take the input strings and output:
- Pathogenicity probability and classification status (Pathogenic / Benign)
- Potential therapeutic candidates generated using my algorithm "ReGen"
- 90.0% ROC-AUC
- 89.7% PR-AUC
- 82.0% F1
- 80.6% Pathogenic F1
- Backend: Python / FastAPI, deployed on Hugging Face Spaces (Docker)
- ML: Scikit-learn, XGBoost
- Algorithm logic and feature engineering pipeline source code in GEM / PRISM (see GitHub profile)
- Frontend: React, CSS, deployed on Vercel
- Derived from GEM and PRISM (see GitHub). Built to explore bioinformatics + ML integration and its potential in the future of medicine.
This project is intended for educational and research purposes only. This is not a medical tool and should not be used for clinical decision-making.